Functional neuroanatomy of spatial sound processing in Alzheimer’s disease

Hannah L Golden1, Jennifer L Agustus1, Jennifer M Nicholas1,2, Jonathan M Schott1,

Sebastian J Crutch1, Laura Mancini3,4, Jason D Warren1

1 Dementia Research Centre, UCL Institute of Neurology, University College London2 London School of Hygiene and Tropical Medicine, University of London3 Neuroradiological Academic Unit, Department of Brain Repair and Rehabilitation, UCL Institute of Neurology,

University4 Lysholm Department of Neuroradiology, National Hospital for Neurology and Neurosurgery.

London, United Kingdom

Correspondence to: Prof Jason WarrenDementia Research CentreUCL Institute of NeurologyUniversity College LondonLondon WC1N 3BGEmail: jason.warren@ucl.ac.ukTel: +44 [0]203 448 4773 Fax: +44 [0]203 448 3104

Running title: fMRI of auditory spatial processing in Alzheimer’s disease

Key words: Alzheimer’s disease; dementia; fMRI; auditory space; auditory scene analysis

Word count: Abstract 173, Main text 6318

Corresponding author: Jason Warren 1

Abstract

Deficits of auditory scene analysis accompany Alzheimer’s disease (AD). However the functional

neuroanatomy of spatial sound processing has not been defined in AD. We addressed this using a ‘sparse’

fMRI virtual auditory spatial paradigm in 14 patients with typical AD in relation to 16 healthy age-matched

individuals. Sound stimulus sequences discretely varied perceived spatial location and pitch of the sound

source in a factorial design. AD was associated with loss of differentiated cortical profiles of auditory

location and pitch processing at the prescribed threshold, and significant group differences were identified

for processing auditory spatial variation in posterior cingulate cortex (controls > AD) and the interaction of

pitch and spatial variation in posterior insula (AD > controls). These findings build on emerging evidence for

altered brain mechanisms of auditory scene analysis and suggest complex dysfunction of network hubs

governing the interface of internal milieu and external environment in AD. Auditory spatial processing may

be a sensitive probe of this interface, and contribute to characterisation of brain network failure in AD and

other neurodegenerative syndromes.

1 Introduction

‘Auditory scene analysis’, the process by which we make sense of our auditory environment (Bregman

1990) entails demanding neural computations that are performed automatically and efficiently by the

normal brain. Auditory scene analysis entails the disambiguation and tracking of sound sources in space

and over time, and has been shown to engage brain mechanisms in auditory association cortex in the

posterior superior temporal lobe and its connections (Bushara et al. 1999; Weeks et al. 1999; Alain et al.

2001, 2008; Warren and Griffiths 2003; Brunetti et al. 2005, 2008; Zimmer et al. 2006; Altmann et al. 2008).

This previous evidence supports a dual organisation of dorsally and ventrally directed human cortical

processing streams respectively mediating sound localisation and identification, and broadly analogous to

the ‘what – where’ dichotomy held to underpin visual object processing. The dorsal auditory stream via its

inferior parietal and premotor projections is involved in preparing behavioural responses to sounds

(Bushara et al. 1999; Weeks et al. 1999; Alain et al. 2001, 2008; Warren et al. 2005; Zimmer et al. 2006) .

2

However, auditory scene analysis is likely to involve additional cortical regions: in particular, the posterior

medial cortical region (comprising posterior cingulate, precuneus and retrosplenial cortex: Leech and Sharp

2014) has been implicated in orienting responses to auditory spatial stimuli (Bushara et al. 1999; Mayer et

al. 2006, 2007; Zündorf et al. 2013) while insula may be engaged in processing aspects of auditory motion

or integrating spatial with other sound characteristics (Griffiths et al. 1994; Lewis et al. 2000; Altmann et al.

2008). Furthermore, the analaysis of natural auditory scenes generally entails sumultaneous processing of

spatial location and identity properties of sound sources in the environment (Bregman 1990).

Recent studies have highlighted the relationship between peripheral hearing function, cognitive

performance and regional brain atrophy (Lin et al. 2011, 2014). However, in addition to any peripheral

hearing effect, the distributed, complex neural computations of auditory scene analysis are likely to be

particularly vulnerable to the cortical pathology of Alzheimer’s disease (AD). Clinical experience suggests

that patients with AD often have difficulty deciphering auditory information in busy acoustic environments

(Golden, Nicholas, et al., 2015). AD has been shown to impair various processes underpinning the analysis

of auditory scenes, including segregation and binding of sound streams (Goll et al. 2012; Golden, Agustus,

et al., 2015), perception of sound location and motion (Kurylo et al. 1993; Golden, Nicholas, et al. 2015),

dichotic listening and auditory attention (Strouse et al. 1995; Gates et al. 1996, 2008, 2011; Golob et al.

2001, 2009). Furthermore, impaired auditory scene analysis may be a harbinger of AD, manifesting

presymptomatically in carriers of pathogenic mutations causing familial AD (Gates et al. 2011; Golob et

al.2009). Deficits of auditory scene analysis in AD have been correlated with alterations of grey matter

structure and function in posterior lateral and medial temporo-parietal cortices that overlap the substrates

of auditory spatial and pitch pattern analysis identified in the healthy brain (Patterson et al. 2002; Warren

and Griffiths 2003; Brunetti et al. 2005; Goll et al. 2012; Zündorf et al. 2013; Golden, Nicholas, et al. 2015) .

These neuroanatomical correlates include core regions of the so-called ‘default-mode network’: a brain

network linking mesial temporal, lateral parietal and prefrontal regions via a posterior medial cortical hub

zone (Shulman et al. 1997; Raichle et al. 2001; Fransson and Marrelec 2008) that has been identified

3

previously as the principal target of the pathological process in AD (Minoshima et al. 1997; Matsuda 2001;

Scahill et al. 2002; Buckner et al. 2005, 2008; Seeley et al. 2009; Lehmann et al. 2010; Warren et al. 2012).

In earlier work, deactivation of the default-mode network on task engagement was interpreted as evidence

that this network mediates stimulus-independent thought in the resting brain (Shulman et al. 1997; Raichle

et al. 2001). However, the network also participates in active processes such as imagery (Buckner and

Carroll 2007; Buckner et al. 2008; Spreng and Grady 2010; Zvyagintsev et al. 2013) which may relate to the

on-line representation of auditory information. The precise role of the default-mode network in these

processes and more particularly the functional impact of AD on this network (and indeed, on connected

brain regions beyond the putative core network) have not been defined. Previous studies utilising task-

related fMRI in AD have focused on memory (Sperling et al. 2003, 2010; Pihlajamäki and Sperling 2009):

while these studies have shown AD is associated with failure to deactivate the default-mode network

normally during information encoding, it remains unclear whether this is a generic mechanism of AD-

mediated dysfunction that extends to other kinds of information processing in sensory systems. Auditory

scene analysis offers a clinically and anatomically relevant paradigm with which to probe AD-associated

network dysfunction, while fMRI provides a means to assess the functional neuroanatomy of component

cognitive processes and to correlate these with behaviour and with structural network disintegration in AD.

Previous functional neuroimaging studies assessing auditory processing in AD have been chiefly confined to

the domain of memory (e.g. Grossman, Koenig, DeVita, et al. 2003; Grossman, Koenig, Glosser, et al. 2003;

Rémy et al. 2005; Peters et al. 2009; Dhanjal and Wise 2014): these studies have revealed a complex profile

of AD-associated network activity shifts. In previous work we have shown that activation of inferior parietal

cortex is increased during auditory scene analysis (the ‘cocktail party effect’) in patients with AD relative to

healthy individuals (Golden, Agustus, et al. 2015). However, previous functional neuroimaging studies have

not assessed the processing of sounds in space: decoding of spatial cues is fundamental to the analysis of

natural auditory scenes, computationally demanding and deficient in AD (Golden, Nicholas, et al. 2015).

In this study, we used fMRI to assess the processing of sound sources located in space in patients with AD

compared with healthy older individuals. We exploited a virtual acoustic space technique that simulates

4

pinna filtering characteristics (Wightman and Kistler 1989a, 1989b) to manipulate sound source location

and pitch in a common paradigm in the scanner environment. Spatial location and pitch are both key

auditory scene components, used in separating and tracking sound sources and information streams

against the acoustic background (Bregman 1990): while evidence for auditory spatial deficits in AD

continues to be amassed (Kurylo et al. 1993; Golden, Nicholas, et al. 2015), the processing of pitch in AD

may be modulated by context and in particular, whether pitch is varied within an auditory scene (Strouse et

al. 1995; Goll et al. 2011, 2012). Although the existence of separable cortical substrates for processing pitch

and spatial information has been established in the healthy brain (Warren and Griffiths, 2003), the extent of

any such dichotomy in the dysfunctional cortex of AD remains unclear. Moreover, natural auditory scenes

typically entail the joint processing of pitch and spatial information and these may interact (Chen et al.

2007). Accordingly, here we adopted a design in which location and pitch were varied factorially in sound

sequences. In addition, we did not employ an output task during scanning, as our primary interest here was

to capture AD-associated alterations in obligatory, ‘bottom-up’ brain mechanisms of spatial sound analysis,

rather than task effects that might potentially be confounded by ‘top-down’ attentional, mnestic or effort

factors; cognitive performance for processing relevant spatial sound parameters was instead assessed in

post-scan behavioural testing. Based on prior cognitive and neuroanatomical evidence (Goll et al. 2012;

Golden, Nicholas, et al. 2015), we hypothesised that AD would be associated with obligatorily altered

cortical signatures of spatial sound analysis relative to healthy individuals, with loss of normal functional

differentiation for the processing of pitch and spatial sound attributes. More specifically, we hypothesised a

functional neuroanatomical correlate of this AD effect in posterior auditory association and temporo-

parietal regions previously implicated in auditory spatial analysis and converging on the default-mode

network (Lewis et al. 2000; Warren and Griffiths 2003; Goll et al. 2012; Zündorf et al. 2013; Golden,

Nicholas, et al. 2015).

2 Methods

2.1 Participants

5

Fourteen consecutive patients [six female; mean (SD) age = 69.8 (6.3)] fulfilling criteria for typical amnestic

AD (Dubois et al. 2007) and 16 healthy older individuals [nine female; mean (SD) age = 70.1 (5.0)] with no

past history of neurological or psychiatric illness participated. No participant had a history of clinically

significant hearing loss. At the time of participation, 12 AD patients were receiving symptomatic treatment

with an acetylcholinesterase inhibitor and the remaining two were receiving memantine. The clinical

diagnosis in the patient group was corroborated by a comprehensive neuropsychological assessment and

volumetric brain MRI; no patient had radiological evidence of significant cerebrovascular damage.

Demographic, clinical and neuropsychological details for all participants are summarised in Table 1. The

diagnosis of AD was further supported by CSF examination (ratio total tau : beta amyloid1-42 >1 in eight

out of nine cases where CSF data were available).

The study was approved by the local institutional ethics committee and all participants gave written

informed consent in accordance with the guidelines laid down in the Declaration of Helsinki.

2.2 Assessment of peripheral hearing

Peripheral hearing was assessed in all participants using a procedure adapted from a commercial screening

audiometry software package (AUDIO-CDTM®, http://www.digital-recordings.com/audiocd/audio.html).

This peripheral audiometry test was administered via headphones from a notebook computer in a quiet

room; participants were presented with continuous tones at one of five frequencies (500, 1000, 2000,

3000, 4000 Hz) that were initially inaudible and slowly and linearly increased in intensity. The task was to

press a button as soon as the participant was sure that a tone had been detected; this response time was

recorded for offline analysis. Hearing was assessed in each ear in each participant.

2.3 Experimental stimuli and conditions

Experimental stimuli were synthesised digitally in MATLAB 2012a® (The Mathworks, Inc.). A series of delay-

and-add functions were applied to a Gaussian noise waveform to create iterated ripple noise (IRN) (Yost,

1996); this provided a broadband carrier that allowed both manipulation of perceived sound source pitch

and spatial location. Perceived pitch was generated by manipulating the latency of the delay between

6

iterations of composite noise waveforms. Perceived spatial location was generated by convolving with

generic head-related transfer functions (HRTFs) that simulate the filtering effect of the pinna and have been

shown to generate a robust percept of a ‘virtual’ sound source in external space (Wightman and Kistler

1989a, 1989b). Five HRTF-specific versions of the stimulus set were created, allowing approximate

matching of the corresponding generic HRTF to an individual participant’s gender and height (see Table S1

in Supplementary Material on-line). All sounds were synthesised with fixed passband 500–5000Hz with

20msec onset–offset ramps to eliminate click artefacts.

The experimental paradigm was adapted from previous work in the healthy young adult brain (Warren and

Griffiths 2003). Five experimental conditions were created for presentation in the scanner, as schematised

in Figure 1 (sound examples are available as Supplementary Material on-line): i) pitch fixed, spatial location

fixed (PfSf); ii) pitch changing, spatial location fixed (PcSf); iii) pitch fixed, spatial location changing (PfSc); iv)

pitch changing, spatial location changing (PcSc); and v) silence. To create the sound conditions, individual

IRN elements of duration 300 milliseconds were concatenated with inter-sound pauses of duration 75

milliseconds to generate sound sequences each containing 21 elements with overall duration 7.8 seconds.

For a given trial (sound sequence), pitch was either fixed or varied randomly between elements of the

sequence with values 70, 85, 100, 115, 130 or 145 Hz, not corresponding to intervals in western music; and

spatial location was either fixed with starting position -90, 0, 90 or 180 degrees or randomly varied with

spatial step size and direction ±30, 40 or 50 degrees in azimuth, such that the initial and final elements

were always identical. This generated a percept of a sound source with constant or randomly varying pitch

that either repeated at the same spatial location or at varying discrete locations around the head.

2.3.1 Stimulus presentation

Stimulus trials were presented from a notebook computer running the Cogent v1.32 extension of MATLAB

(http://www.vislab.ucl.ac.uk/cogent_2000.php), each triggered by the MR scanner on completion of the

previous image acquisition in a ‘sparse’ acquisition protocol. Sounds were delivered binaurally via

electrodynamic headphones (http://www.mr-confon.de/) at a comfortable listening level (at least 70 dB)

that was fixed for all participants; two identical scanning runs were administered, each comprising 16 trials

7

for each sound condition plus eight silence trials, yielding a total of 144 trials for the experiment.

Participants were instructed to listen to the sound stimuli with their eyes open; there was no in-scanner

output task or visual fixation constraint and no behavioural responses were collected.

2.3.2 Brain image acquisition

Brain images were acquired on a 3Tesla Trio MRI scanner (Siemens, Erlangen, Germany) using a 12-channel

RF receive head coil. For each of the two functional runs, 74 single-shot gradient-echo planar image (EPI)

volumes were acquired each with 48 oblique transverse slices covering the whole brain (slice thickness

2mm, inter-slice gap 1mm and 3mm in-plane resolution, TR/TE 70/30ms, echo spacing 0.5ms, matrix size 64

x 64 pixels, FoV 192 x 192mm, phase encoding (PE) direction anterior-posterior). A slice tilt of -30o (T>C), z-

shim gradient moment of +0.6 mT/m*ms and positive PE gradient polarity were used to minimise

susceptibility-related loss of signal and blood-oxygen-level-dependent functional sensitivity in the temporal

lobes, following optimisation procedures described previously (Weiskopf et al. 2006). Sparse-sampling EPI

acquisition with repetition time 11.36s (corresponding to an inter-scan gap of 8 seconds) was used to

reduce any interaction between scanner acoustic noise and auditory stimulus presentations. The initial two

brain volumes in each run were performed to allow equilibrium of longitudinal T1 magnetisation and

discarded from further analysis. A B0 field-map was acquired (TR = 688ms; TE1 = 4.92ms, TE2 = 7.38ms,

3x3x3mm resolution, no interslice gap; matrix size = 80 x 80 pixels; FoV = 192 x 192mm; phase encoding

direction = A-P) to allow post-processing geometric distortion corrections of EPI data due to B0 field

inhomogeneities.

A volumetric brain MR image was also obtained in each participant to allow coregistration of structural with

functional neuroanatomical data. The head coil was switched to a 32 channel RF receiver head coil

(Siemens, Erlangen, Germany). T1-weighted volumetric images were obtained using a sagittal 3-D

magnetization prepared rapid gradient echo sequence (TR = 2200ms; TE = 2.9ms; matrix size 256 × 256

pixels, voxel size of 1.1 × 1.1 × 1.1 mm).

2.3.3 Post-scan behavioural testing

8

Following the scanning session, each participant’s ability to perceive the key experimental parameters of

the fMRI experiment was assessed using alternative forced choice psychoacoustic procedures that assessed

pitch change detection and auditory spatial location change detection. Twenty stimuli from the scanning

session were used (five for each of the four sound conditions). For the spatial subtest, the task on each trial

was to decide whether the sounds were fixed in position or changing between positions. For the pitch

subtest, the task on each trial was to decide whether the sounds were fixed or changing in pitch. It was

established that all participants understood the tasks prior to commencing the tests; during the tests, no

feedback about performance was given and no time limits were imposed. All responses were recorded for

off-line analysis.

2.4 Analysis of fMRI data

Brain image data were analysed using statistical parametric mapping software (SPM8;

http://www.fil.ion.ucl.ac.uk/spm). In initial image pre-processing, the EPI functional series for each

participant was realigned using the first image as a reference, and images were unwarped incorporating

field-map distortion information (Hutton et al. 2002). The DARTEL toolbox (Ashburner 2007) was used to

spatially normalise all individual functional images to a group mean template image in Montreal

Neurological Institute (MNI) standard stereotactic space; to construct this group brain template, each

individual’s T1 weighted MR image was first coregistered to their EPI series and segmented using DARTEL

tools (New Segment) and this segment was then used to estimate a group template that was aligned to

MNI space. Functional images were smoothed using a 6mm full-width-at-half-maximum Gaussian

smoothing kernel. For the purpose of rendering statistical parametric functional maps, a study-specific

mean structural brain image template was created by warping all bias-corrected native space whole-brain

images to the final DARTEL template and calculating the average of the warped brain images.

Pre-processed functional images were entered into a first-level design matrix incorporating the five

experimental conditions modelled as separate regressors convolved with the standard haemodynamic

response function, and also including six head movement regressors generated from the realignment

process. For each participant, first-level t-test contrast images were generated for the main effects of

9

auditory stimulation [(PfSf + PfSc + PcSf + PcSc) – silence], changing pitch [(PcSc + PcSf) – (PfSc + PfSf)],

changing spatial location [(PcSc + PfSc) – (PcSf + PfSf)] and the interaction of these effects [(PcSc – PcSf) –

(PfSc – PfSf)]. Both ‘forward’ and ‘reverse’ contrasts were assessed in each case. Contrast images for each

participant were entered into a second-level random-effects analysis in which effects within each

experimental group and between the healthy control and AD groups were assessed using voxel-wise t-test

contrasts.

Contrasts were assessed at a peak-level significance threshold p < 0.05 after family-wise error (FWE)

correction for multiple voxel-wise comparisons within neuroanatomical regions of interest in each cerebral

hemisphere pre-specified by our prior anatomical hypotheses. These anatomical small volumes comprised

anterior superior temporal gyrus regions previously implicated in processing pitch patterns (Patterson et al.

2002; Warren and Griffiths 2003; Arnott et al. 2004) and spatial characteristics of auditory scenes:

temporoparietal junction (posterior temporal lobe and angular gyrus), posterior medial cortex (posterior

cingulate, precuneus, retrosplenial cortex) and insula (Griffiths et al. 1994; Lewis et al. 2000; Warren and

Griffiths 2003; Arnott et al. 2004; Brunetti et al. 2005, 2008; Shomstein and Yantis 2006; Zündorf et al.

2013). A region that combined anterior and posterior superior temporal gyri to encompass primary and

association auditory cortex was used for the contrast assessing all sound activation. Anatomical regions

were derived from Oxford-Harvard cortical (Desikan et al. 2006) and Jülich histological (Eickhoff et al. 2005)

maps via FSLview (Jenkinson et al., 2012) and further edited in MRICron®

(http://www.mccauslandcenter.sc.edu/mricro/mricron/) to conform to the study-specific template brain

image; the regions are presented in Figure S1 in Supplementary Material on-line.

2.5 Analysis of structural MRI data

Structural brain images were compared between the patient and healthy control groups in a voxel-based

morphometric (VBM) analysis to obtain an AD-associated regional atrophy map: normalisation,

segmentation and modulation of grey and white matter images were performed using default parameter

settings in SPM8, with a Gaussian smoothing kernel of 6mm full-width-at-half-maximum. Groups were

compared using voxel-wise two-sample t-tests, including covariates of age, gender, and total intracranial

10

volume. Statistical parametric maps of brain atrophy were thresholded leniently (p < 0.01 uncorrected for

multiple voxel-wise comparisons over the whole brain volume) in order to capture any significant grey

matter structural changes in relation to functional activation profiles from the fMRI analysis.

2.6 Analysis of demographic and behavioural data

Demographic data were compared between the healthy control and AD groups using two sample t-tests

(gender differences were assessed using Pearson’s chi-square test of distribution); neuropsychological data

were compared using non-parametric Wilcoxon rank-sum tests. Tone detection thresholds on audiometry

screening and performance on post-scan behavioural tasks on experimental stimuli were analysed using

linear regression models with clustered, robust standard error due to non-equal variance between groups.

In the audiometry analysis, the main effect of patient group was assessed whilst controlling for age and

frequency type, as well as assessing for any interaction between group and frequency. In the analysis of

post-scan behavioural data, a robust, cluster-adjusted regression model was utilised to test for the main

effects of disease and behavioural task on proportion of correct answers while also testing for any

interaction between these two factors. Wald tests were used to further assess effects of interactions and

specific hypotheses. Spearman’s correlations were performed to assess any association between peak

activation for specific contrast beta weights in the fMRI analysis and d-prime scores for performance on the

out of scanner behavioural tasks for each participant group.

3 Results

3.1 General participant characteristics

Results of the analysis of demographic and behavioural data are summarised in Table 1. The patient and

healthy control groups were well matched for age (t (28) = 0.13, p = 0.89) and gender distribution (χ2(1) = 0.15,

p = 0.70), however the control group had on average significantly more years of education (t (28) = 2.57, p =

0.02); years of education was accordingly included as a covariate of no interest in subsequent analyses of

behavioural data. As anticipated, the AD group performed significantly worse than the healthy control

group on a range of neuropsychological measures; referenced to normative data for this age group, AD

11

patients showed particularly severe deficits of episodic memory, executive function, naming and

visuospatial working memory. Tone detection thresholds on audiometry did not differ between the patient

and healthy control groups (beta = 170, p = 0.94, CI -4198 to 4540), nor was there any significant

interaction between group and sound frequency (F (4,29) = 1.11, p = 0.37); accordingly, peripheral hearing

function was not considered further as a factor in analyses.

3.2 Post-scan behavioural data

Group performance data for the post-scan behavioural tests are presented in Table 1. The AD group

performed significantly worse than the healthy control group on both the pitch and spatial tasks (beta = -

3.32, p = 0.006, CI -5.60 to -1.03); scores did not differ significantly between task type (beta = -0.75, p =

0.193, CI -1.90 to 0.40) and there was no significant interaction between group and test type (F (1,29) = 0.90, p

= 0.35). Eight individuals with AD on the spatial task and three on the pitch task performed below the range

of the healthy control group.

3.3 Structural neuroanatomical data

Comparison of the AD and healthy control groups in the VBM analysis revealed the anticipated profile of

AD-associated regional grey matter atrophy involving hippocampi, temporal, temporoparietal and posterior

medial cortices. Statistical parametric maps are presented in Figure 2 and further details about regional

atrophy profiles with local maxima of grey matter loss are presented in Supplementary Table S2 in

Supplementary Material on-line.

3.4 Functional neuroanatomical data

Statistical parametric maps of significant activation for contrasts of interest are presented in Figure 3 and in

Figure S2 in Supplementary Material on-line; significant local maxima are summarised in Table 2 (additional

activations observed at a more lenient significance threshold p<0.001 uncorrected for multiple comparisons

over the whole brain are presented in Table S3 in Supplementary Material on-line). Auditory stimulation

(the contrast of all sound conditions over silence) produced as anticipated extensive bilateral activation of

Heschl’s gyrus and superior temporal gyrus, in both the healthy control and AD groups (see Figure S2). Pitch

variation (changing over fixed pitch) produced activation of right anterior superior temporal gyrus and 12

sulcus in the healthy control group but no activation in the AD group at the prescribed threshold (activation

was observed for the AD group in posterior superior temporal cortex at a relaxed uncorrected threshold;

Figure 3 and Table S3). Auditory spatial variation (changing over fixed sound location) produced bilateral

activation of posterior superior temporal gyrus, planum temporale and posterior cingulate cortex in the

healthy control group but no activation in the AD group at the prescribed threshold. No significant

activations were identified for the ‘reverse’ contrasts of fixed over changing pitch or fixed over changing

spatial location. The interaction of spatial and pitch variation elicited significant activation in left anterior

superior temporal cortex in the healthy control group and significant activation in right posterior insula in

the AD group.

When the AD and healthy control groups were compared directly, the effect of auditory spatial variation

was significantly greater in the healthy control group than the AD group in posterior cingulate cortex

(Figure 3). Post hoc analysis of condition beta weights revealed that this group-wise interaction was driven

by significantly higher beta values for conditions with changing versus fixed auditory spatial location

(greater deactivation in conditions with fixed auditory spatial location) in posterior cingulate in the healthy

control group. The interaction of auditory spatial and pitch variation produced significantly greater

activation of right posterior insula in the AD group versus the healthy control group; post hoc analysis of

condition beta weights for this interaction revealed no significant pair-wise group or condition differences

but rather, mirror beta profiles in the two groups (the AD group showed less activation in conditions where

pitch or auditory spatial change occurred in isolation than in conditions where pitch and auditory spatial

location were both fixed or changing simultaneously, while the healthy control group showed the reverse

pattern).

The healthy control group showed a significant inverse correlation between peak activation in posterior

cingulate cortex and d-prime for the auditory spatial task (r (s) = -0.55, p = 0.03) but no significant

correlations between peak activation in insula and d-prime for either task. The AD group showed no

significant correlations between peak regional activations and d-prime for either task (spatial task in

13

posterior cingulate, r(s) = 0.34, p = 0.23; spatial task in right posterior insula r (s) = 0.03, p = 0.93; pitch task in

posterior insula, r(s) = -0.38, p = 0.18).

4 Discussion

Here we have shown that functional neuroanatomical mechanisms for processing spatial sounds are altered

in AD compared to the healthy older brain. Elementary sound encoding (the effect of any auditory

stimulation compared with silence) produced similar activation in patients with AD and in healthy older

individuals, indicating that AD targets higher order processing of sound attributes. In the older control

group, the processing of sequential pitch variation activated anterior superior temporal cortex, consistent

with previous evidence for pitch pattern analysis in the healthy brain (Patterson et al. 2002; Warren and

Griffiths 2003). While this activation profile was not observed at the prescribed threshold in the patients

with AD, the experimental groups did not differ significantly in the processing of pitch variation per se. In

contrast, the groups did show significantly different activation profiles in response to changing sound

location in posterior cingulate cortex. This was driven chiefly by failure of the normal deactivation of

posterior cingulate cortex in the fixed auditory spatial location conditions in AD group (Figure 3); AD was

associated with loss of functional differentiation of posterior cingulate responses that was evident in

healthy older individuals. Unlike the healthy control group, the AD group showed an interaction between

pitch and spatial sequence processing in posterior insula and this group difference was also significant. The

form of this interaction was complex and driven by mirror profiles of activation in the AD and healthy

control groups (Figure 3): the normal profile of enhanced activation shown by controls in conditions with

congruent compared with incongruent pitch and spatial variation was reversed in the AD group.

Furthermore, functional neuroanatomical differences between the AD and healthy older control groups

extended (particularly in the case of the insular interaction effect) beyond the zone of disease-associated

grey matter atrophy as characterised in a parallel structural neuroanatomical comparison between the

groups (Figure 2).

14

Taken together, these findings suggest that AD is associated with specific functional alterations in a brain

network engaged in processing spatial sounds. This study builds on previous evidence in the healthy brain

demonstrating that posterior medial cortex is engaged during analysis of both spatial and nonspatial

information in auditory scenes (Bushara et al. 1999; Mayer et al. 2006, 2007; Wong et al. 2008, 2009;

Zündorf et al. 2013). More particularly, the present work corroborates other evidence for dysfunction of

brain mechanisms that mediate aspects of auditory scene analysis including auditory source localisation in

AD (Kurylo et al. 1993; Strouse et al. 1995; Gates et al. 1996, 2008, 2011; Golob et al. 2001, 2009; Goll et al.

2012; Golden, Agustus, et al. 2015; Golden, Nicholas, et al. 2015). Previous studies of patients with AD have

demonstrated structural grey matter correlates of sound stream disambiguation in posterior cingulate

cortex and auditory spatial discrimination in precuneus (Goll et al. 2012; Golden, Nicholas, et al. 2015).

Posterior medial cortical regions generally work in concert to mediate various aspects of self-awareness

and directed attention (Vogt and Laureys 2005; Leech and Sharp 2014). While the precise roles of these

cortices in auditory scene analysis have not been defined, posterior medial cortex may be preferentially

involved in reorienting of attention between locations in egocentric space (Mayer et al. 2006, 2007;

Shomstein and Yantis 2006) or (during passive spatial listening, as here) implicit tracking of a sound source

over a series of location shifts around the head. This may reflect a broader role of this cortical region in the

cognition of spatial navigation (Miller et al. 2014). Moreover, posterior cingulate cortex is a hub zone of the

putative default-mode network implicated as core to the pathogenesis of AD (Minoshima et al. 1997;

Buckner et al. 2005; Fransson and Marrelec 2008; Lehmann et al. 2010; Warren et al. 2012).

The role of insula in processing auditory information continues to be defined. Unlike posterior cingulate

cortex insula is not a core DMN component but it is likely to act as a multimodal hub that integrates body

state information with incoming sensory traffic from the external environment: as such, this region is well

placed to link DMN with brain networks that evaluate sensory stimuli and programme behavioural

responses, in particular the anterior fronto-insular ‘salience network’ (Seeley et al. 2009; Zhou and Seeley

2014). Previous work has implicated insula cortex in the analysis of sound movement particularly motion

relative to self (Griffiths et al. 1994, 1997; Lewis et al. 2000), however this multimodal region has functional

15

subdivisions and a range of potentially relevant functions that have yet to be fully defined (Bamiou et al.

2003): these include fine-grained analysis of auditory timing cues (Bamiou et al. 2006) and the modulation

of spatial by-nonspatial auditory object features (Altmann et al. 2008). Insular activity is sensitive to

cognitive load in the processing of musical and other sound patterns (Altmann et al. 2008; Nan et al. 2008)

and to the detection of changes across sensory modalities (Downar et al. 2000): considered together with

evidence that insula and its connections to DMN are affected relatively early in the course of AD (Xie et al.

2012), it is therefore plausible that the interaction of spatial and pitch pattern processing here should

engage insular cortex in AD but not in the healthy older brain. Activity in this region was not correlated with

performance on post-scan spatial or pitch discrimination tasks in the present patient cohort, suggesting

that this engagement of insula in AD did not fulfil any compensatory role during auditory scene analysis.

Insular involvement here might in principle reflect differential engagement of multimodal regions during

processing of computationally demanding sensory traffic (for example, calibration of a stable pitch or

spatial template while the other parameter is changing); alternatively, it might represent an entirely

aberrant activation profile produced by AD. One potentially unifying interpretation of the present findings

might invoke dysfunctional coupling between posterior cingulate and insular cortex in AD, leading to

impaired ability to update mental representations of a sound source with shifting spatial or pitch

trajectories: this would be consistent with a role for posterior cingulate cortex in tuning brain network

activity between internally and externally directed cognitive operations (Leech and Sharp 2014).

This profile of cortical dysfunction in AD is unlikely to reflect simply attenuation of activity due to

pathological grey matter loss. Inspection of condition effect sizes in the present healthy control and AD

groups (Figure 3) reveals complex profiles of bidirectional activity shifts in AD patients relative to healthy

older individuals. In particular, posterior cingulate cortex in AD patients did not show the normal pattern of

reduced activation in response to sounds with fixed versus changing spatial location. Whilst activity shifts

are more difficult to interpret in the absence of an output task, this pattern in AD is consistent with failure

to deactivate posterior cingulate cortex normally. In the healthy brain, deactivation of posterior cingulate

might play a crucial modulatory or permissive role in bringing other brain areas on-line during analysis of

16

spatial sounds: this interpretation is in-line with the present data (Figure 3) which further suggest that AD

leads to a loss of the normal inverse correlation between posterior cingulate activity and auditory spatial

perceptual performance (as indexed here by the out-of-scanner behavioural task). An analogous failure to

modulate activity in posterior medial cortex has been linked previously to impaired memory performance

in AD (Sperling et al. 2003, 2010; Celone et al. 2006; Pihlajamäki and DePeau 2008; Pihlajamäki and Sperling

2009) and may consititute a generic mechanism of AD-associated default-mode dysfunction. It is likely that

dysfunction of this key hub region is modulated by connectivity with other brain regions and by tasks

engaging auditory attention (Kamourieh et al., 2015). More fundamentally, it remains unclear to what

extent altered BOLD signal responses may reflect the effects of AD or acetylcholinesterase inhibitor

treatment on regional cerebral haemodynamic responses (Rombouts et al. 2005; Bentley et al. 2008;

Thiyagesh et al. 2010).

Though this study was not primarily designed to elucidate brain mechanisms of auditory ‘what’ and ‘where’

processing, our findings support a functional neuroanatomical dichotomy for processing spatial and

nonspatial auditory information in healthy older controls, albeit with some potential for interaction

between these dimensions. Relative to healthy controls, the AD group showed comparable behavioural

deficits in processing both pitch changes and location changes. Whereas pitch processing has been found to

be relatively preserved relative to spatial processing in previous neuropsychological studies of AD (Kurylo et

al. 1993; Strouse et al. 1995; Goll et al. 2012; Golden, Nicholas, et al. 2015), the stimuli used here depart

from previous work in requiring conjoint processing of pitch changes in the presence of simultaneous

spatial cues and over extended sound sequences. It might therefore be argued that the present stimuli

more closely reflect the increased task demands imposed by natural auditory scenes, in which pitch

information must be extracted (as here) from sounds in space. The lack of functional neuroanatomical

differentiation between groups for the pitch processing contrast here therefore appears somewhat

paradoxical, but might be attributable to several factors. Care is needed, firstly, in interpreting null effects

in fMRI analyses since these at least in part reflect statistical thresholding (using a more relaxed whole

brain threshold, activation was evident in the AD group for the pitch contrast though not the spatial

17

contrast: see Figure 3 and Table S3). In addition, the out-of-scanner behavioural tasks here were not

intended to provide a detailed stratification of pitch and spatial processing impairments, which are more

fully delineated using customised, graded difficulty stimuli (Golden, Nicholas et al., 2015). Moreover,

behavioural deficits need not have a discrete regional neuroanatomical mapping: the pitch deficit in the AD

group relative to healthy controls might, for example, arise from altered network connectivity, which was

not captured here. The lack of significant within-group fMRI signatures of pitch and spatial change

processing in the present AD group at the prescribed threshold occurred despite a normal response to

auditory stimulation per se: rather than some generalised failure of auditory cortical processing, AD may

predominantly affect the processing of higher-order sound attributes. The lack of significantly differentiable

cortical signatures of pitch and spatial processing and their abnormal interaction in posterior insula

together argue for loss of selectivity of auditory cortical mechanisms in AD. The present data extend earlier

work showing that the pathological process in AD targets cortical mechanisms of auditory scene analysis

(Goll et al. 2012; Golden, Agustus, et al. 2015; Golden, Nicholas, et al. 2015) and more broadly, align with

other evidence that the functional integrity of the default mode network and interacting cortical networks

is disrupted in AD (Dennis and Thompson 2014).

This study has certain limitations that suggest directions for future work. Case numbers here were relatively

small; the findings should be substantiated in larger cohorts representing AD phenotypic variants, which

may have distinct auditory spatial signatures (Golden, Nicholas, et al. 2015) as well as non-AD dementias.

Deficits of auditory scene analysis may be early markers of AD (Gates et al. 2002, 2011; Golob et al. 2009):

this should be further assessed in longitudinal studies with fMRI correlation, ideally including

presymptomatic individuals with genetic AD. The present passive listening paradigm was designed to

address mechanisms of obligatory perceptual analysis. These mechanisms are likely to be modulated by

output task, memory and attentional demands (Warren et al. 2005; Kamourieh et al. 2015) and by

mechanisms for coding behavioural stimulus salience that may also be altered in AD (Fletcher et al. 2015):

such factors should be investigated explicitly. Related to this, the processing of spatial sounds should be

assessed under more ecological conditions requiring integration of multimodal cues. Besides anatomical

18

mapping, functional network connectivity alterations may capture additional disease effects and should be

investigated directly (for example, using graph theoretical techniques). From a more basic physiological

perspective, interpretation of fMRI studies in AD will require elucidation of the impact of disease and drugs

modulating cholinergic function on cerebral haemodynamic responses, defined using continuous sampling

of the BOLD signal rather than the sparse ‘snapshots’ captured in the present acquisition protocol. Taking

these limitations into account, this study consolidates a growing body of work suggesting that auditory

scene analysis may be a sensitive probe of brain network disintegration in AD (Goll et al. 2012; Golden,

Agustus, et al. 2015; Golden, Nicholas, et al. 2015). Tracking sound sources in space requires updating of an

internal sensory image by incoming sensory information and precise dynamic coding of sensory signals:

neural operations that are likely to be peculiarly vulnerable to the anatomical topography of AD (Vogt and

Laureys 2005; Leech and Sharp 2014) and to the effects of neurodegenerative pathology on essential

electrophysiological properties of cortical neurons (Ahveninen et al. 2014). Future work could test these

ideas directly by comparing large-scale brain network interactions in AD and diseases (such as the

frontotemporal lobar degenerations) with distinct network signatures (Zhou and Seeley 2014); and by

manipulating spatial and nonspatial attributes of more complex, naturalistic auditory ‘scenes’, such as

music.

5 Acknowledgments

We are grateful to all participants for their involvement, to all the radiographers at the National Hospital for

Neurology and Neurosurgery for assistance with MRI scanning, and to Professor Gill Livingston for referring

research patients. This work was supported by the Wellcome Trust (091673/Z/10/Z to JDW) and Alzheimer

Research UK (ART-PhD2011-10 to HLG; ART-SRF2010-3 to SJC), the Economic and Social Research Council

(ES/K006711/1) and the National Institute of Health Research Queen Square Dementia Biomedical Research

Unit (CBRC 161). The authors report no conflicts of interest.

19

6 References

Ahveninen J, Kopčo N, Jääskeläinen IP. 2014. Psychophysics and neuronal bases of sound localization in humans. Hear Res. 307:86–97.

Alain C, Arnott SR, Hevenor S, Graham S, Grady CL. 2001. “What” and “where” in the human auditory system. Proc Natl Acad Sci U S A. 98:12301–12306.

Alain C, He Y, Grady C. 2008. The contribution of the inferior parietal lobe to auditory spatial working memory. J Cogn Neurosci. 20:285–295.

Altmann CF, Henning M, Döring MK, Kaiser J. 2008. Effects of feature-selective attention on auditory pattern and location processing. Neuroimage. 41:69–79.

Arnott SR, Binns MA, Grady CL, Alain C. 2004. Assessing the auditory dual-pathway model in humans. Neuroimage. 22:401–408.

Ashburner J. 2007. A fast diffeomorphic image registration algorithm. Neuroimage. 38:95–113.Bamiou DE, Musiek FE, Luxon LM. 2003. The insula (Island of Reil) and its role in auditory processing:

Literature review. Brain Res Rev. 42:143–154.Bamiou D-E, Musiek FE, Stow I, Stevens J, Cipolotti L, Brown MM, Luxon LM. 2006. Auditory temporal

processing deficits in patients with insular stroke. Neurology. 67:614–619.Bentley P, Driver J, Dolan RJ. 2008. Cholinesterase inhibition modulates visual and attentional brain

responses in Alzheimer’s disease and health. Brain. 131:409–424.Bregman AS. 1990. Auditory Scene Analysis: The Perceptual Organization of Sound. MIT Press.Brunetti M, Belardinelli P, Caulo M, Del Gratta C, Della Penna S, Ferretti A, Lucci G, Moretti A, Pizzella V,

Tartaro A, Torquati K, Belardinelli MO, Romani GL. 2005. Human brain activation during passive listening to sounds from different locations: An fMRI and MEG study. Hum Brain Mapp. 26:251–261.

Brunetti M, Della Penna S, Ferretti a., Del Gratta C, Cianflone F, Belardinelli P, Caulo M, Pizzella V, Olivetti Belardinelli M, Romani GL. 2008. A Frontoparietal network for spatial attention reorienting in the auditory domain: A human fMRI/MEG study of functional and temporal dynamics. Cereb Cortex. 18:1139–1147.

Buckner RL, Andrews-Hanna JR, Schacter DL. 2008. The brain’s default network: anatomy, function, and relevance to disease. Ann N Y Acad Sci. 1124:1–38.

Buckner RL, Carroll DC. 2007. Self-projection and the brain. Trends Cogn Sci. 11:49–57.Buckner RL, Snyder AZ, Shannon BJ, LaRossa G, Sachs R, Fotenos AF, Sheline YI, Klunk WE, Mathis CA,

Morris JC, Mintun MA. 2005. Molecular, structural, and functional characterization of Alzheimer’s disease: evidence for a relationship between default activity, amyloid, and memory. J Neurosci. 25:7709–7717.

Bushara KO, Weeks RA, Ishii K, Catalan MJ, Tian B, Rauschecker JP, Hallett M. 1999. Modality-specific frontal and parietal areas for auditory and visual spatial localization in humans. Nat Neurosci. 2:759–766.

Celone KA, Calhoun VD, Dickerson BC, Atri A, Chua EF, Miller SL, DePeau K, Rentz DM, Selkoe DJ, Blacker D, Albert MS, Sperling RA. 2006. Alterations in memory networks in mild cognitive impairment and Alzheimer’s disease: an independent component analysis. J Neurosci. 26:10222–10231.

Chen Q, Zhang M, Zhou X. 2007. Interaction between location- and frequency-based inhibition of return in human auditory system. Exp Brain Res. 176:630–640.

Delis DC, Kaplan E, Kramer JH. 2001. Delis-Kaplan executive function system. San Antonio, TX: The Psychological Corporation.

Dennis EL, Thompson PM. 2014. Functional brain connectivity using fMRI in aging and Alzheimer’s disease. Neuropsychol Rev. 24:49–62.

Desikan RS, Ségonne F, Fischl B, Quinn BT, Dickerson BC, Blacker D, Buckner RL, Dale AM, Maguire RP, Hyman BT, Albert MS, Killiany RJ. 2006. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage. 31:968–980.

Dhanjal NS, Wise RJS. 2014. Frontoparietal cognitive control of verbal memory recall in Alzheimer’s disease. Ann Neurol. 76:241–251.

20

Downar J, Crawley AP, Mikulis DJ, Davis KD. 2000. A multimodal cortical network for the detection of changes in the sensory environment. Nat Neurosci. 3:277–283.

Dubois B, Feldman HH, Jacova C, Dekosky ST, Barberger-Gateau P, Cummings J, Delacourte A, Galasko D, Gauthier S, Jicha G, Meguro K, O’brien J, Pasquier F, Robert P, Rossor M, Salloway S, Stern Y, Visser PJ, Scheltens P. 2007. Research criteria for the diagnosis of Alzheimer’s disease: revising the NINCDS-ADRDA criteria. Lancet Neurol. 6:734–746.

Dunn LM, Dunn PQ, Whetton C. 1982. British Picture Vocabulary Scale. Windsor: NFER-Nelson.Eickhoff SB, Stephan KE, Mohlberg H, Grefkes C, Fink GR, Amunts K, Zilles K. 2005. A new SPM toolbox for

combining probabilistic cytoarchitectonic maps and functional imaging data. Neuroimage. 25:1325–1335.

Fletcher PD, Nicholas JM, Shakespeare TJ, Downey LE, Golden HL, Agustus JL, Clark CN, Mummery CJ, Schott JM, Crutch SJ, Warren JD. 2015. Dementias show differential physiological responses to salient sounds. Front Behav Neurosci. 9:1–8.

Fransson P, Marrelec G. 2008. The precuneus/posterior cingulate cortex plays a pivotal role in the default mode network: Evidence from a partial correlation network analysis. Neuroimage. 42:1178–1184.

Gates G, Cobb J, Linn R. 1996. Central auditory dysfunction, cognitive dysfunction, and dementia in older people. Arch Otolaryngol Head Neck Surg. 122:151–167.

Gates GA, Anderson ML, Feeney MP, McCurry SM, Larson EB. 2008. Central auditory dysfunction in older persons with memory impairment or Alzheimer dementia. Arch Otolaryngol Head Neck Surg. 134:771–777.

Gates GA, Anderson ML, McCurry SM, Feeney MP, Larson EB. 2011. Central Auditory Dysfunction as a Harbinger of Alzheimer Dementia. Arch Otolaryngol Neck Surg. 137:390–395.

Gates GA, Beiser A, Rees TS, D’Agostino RB, Wolf PA. 2002. Central Auditory Dysfunction May Precede the Onset of Clinical Dementia in People with Probable Alzheimer’s Disease. J Am Geriatr Soc. 50:482–488.

Golden HL, Agustus JL, Goll JC, Downey LE, Mummery CJ, Schott JM, Crutch SJ, Warren JD. 2015. Functional neuroanatomy of auditory scene analysis in Alzheimer’s disease. NeuroImage Clin. 7:699–708.

Golden HL, Nicholas JM, Yong KXX, Downey LE, Schott JM, Mummery CJ, Crutch SJ, Warren JD. 2015. Auditory spatial processing in Alzheimer’s disease. Brain. 138:189–202.

Goll JC, Kim LG, Hailstone JC, Lehmann M, Buckley A, Crutch SJ, Warren JD. 2011. Auditory object cognition in dementia. Neuropsychologia. 49:2755–2765.

Goll JC, Kim LG, Ridgway GR, Hailstone JC, Lehmann M, Buckley AH, Crutch SJ, Warren JD. 2012. Impairments of auditory scene analysis in Alzheimer’s disease. Brain. 135:190–200.

Golob EJ, Miranda GG, Johnson JK, Starr A. 2001. Sensory cortical interactions in aging , mild cognitive impairment , and Alzheimer ’ s disease. Neurobiol Aging. 22:755–763.

Golob EJ, Ringman JM, Irimajiri R, Bright S, Schaffer B, Medina LD, Starr A. 2009. Cortical event-related potentials in preclinical familial Alzheimer disease. Neurology. 73:1649–1655.

Griffiths TD, Bench CJ, Frackowiak RSJ. 1994. Human cortical areas selectively activated by apparent sound movement. Curr Biol. 4:892–895.

Griffiths TD, Rees A, Witton C, Cross PM, Shakir RA, Green GGR. 1997. Spatial and temporal auditory processing deficits following right hemisphere infarction - A psychophysical study. Brain. 120:785–794.

Grossman M, Koenig P, DeVita C, Glosser G, Moore P, Gee J, Detre J, Alsop D. 2003. Neural basis for verb processing in Alzheimer’s disease: an fMRI study. Neuropsychology. 17:658–674.

Grossman M, Koenig P, Glosser G, DeVita C, Moore P, Rhee J, Detre J, Alsop D, Gee J. 2003. Neural basis for semantic memory difficulty in Alzheimer’s disease: an fMRI study. Brain. 126:292–311.

Hutton C, Bork A, Josephs O, Deichmann R, Ashburner J, Turner R. 2002. Image distortion correction in fMRI: A quantitative evaluation. Neuroimage. 16:217–240.

Jackson M, Warrington EK. 1986. Arithmetic skills in patients with unilateral cerebral lesions. Cortex. 22:611–620.

Kamourieh S, Braga RM, Leech R, Newbould RD, Malhotra P, Wise RJS. 2015. Neural Systems Involved When Attending to a Speaker. Cereb Cortex. 1–15.

21

Kurylo D, Corkin S, Allard T, Zatorre R, Growdon J. 1993. Auditory function in Alzheimer’s disease. Neurology. 43:1893–1899.

Leech R, Sharp DJ. 2014. The role of the posterior cingulate cortex in cognition and disease. Brain. 137:12–32.

Lehmann M, Rohrer JD, Clarkson MJ, Ridgway GR, Scahill RI, Modat M, Warren JD, Ourselin S, Barnes J, Rossor MN, Fox NC. 2010. Reduced Cortical Thickness in the Posterior Cingulate Gyrus is Characteristic of Both Typical and Atypical Alzheimer’s Disease. J Alzheimers Dis. 20:587–598.

Lewis JW, Beauchamp MS, DeYoe EA. 2000. A comparison of visual and auditory motion processing in human cerebral cortex. Cereb Cortex. 10:873–888.

Lin F, Metter E, O’Brien R, Resnick S, Zonderman A, Ferrucci L. 2011. Hearing loss and incident dementia. Arch Neurol. 68:214–220.

Lin FR, Ferrucci L, An Y, Goh JO, Doshi J, Metter EJ, Davatzikos C, Kraut MA, Resnick SM. 2014. Association of hearing impairment with brain volume changes in older adults. Neuroimage. 90:84–92.

Matsuda H. 2001. Cerebral blood flow and metabolic abnormalities in Alzheimer’s disease. Ann Nucl Med. 15:85–92.

Mayer AR, Harrington D, Adair JC, Lee R. 2006. The neural networks underlying endogenous auditory covert orienting and reorienting. Neuroimage. 30:938–949.

Mayer AR, Harrington DL, Stephen J, Adair JC, Lee RR. 2007. An event-related fMRI Study of exogenous facilitation and inhibition of return in the auditory modality. J Cogn Neurosci. 19:455–467.

McKenna P, Warrington E. 1983. Graded naming test. Windsor: NFER-Nelson.Miller AMP, Vedder LC, Law LM, Smith DM. 2014. Cues, context, and long-term memory: the role of the

retrosplenial cortex in spatial cognition. Front Hum Neurosci. 8:1–15.Minoshima S, Giordani B, Berent S, Frey KA, Foster NL, Kuhl DE. 1997. Metabolic reduction in the posterior

cingulate cortex in very early Alzheimer’s disease. Ann Neurol. 42:85–94.Nan Y, Knösche TR, Zysset S, Friederici AD, Friedend AD. 2008. Cross-cultural music phrase processing: an

fMRI study. Hum Brain Mapp. 29:312–328.Nelson HE. 1982. National Adult Reading Test. Windsor: NFER-Nelson.Patterson RD, Uppenkamp S, Johnsrude IS, Griffiths TD. 2002. The processing of temporal pitch and melody

information in auditory cortex. Neuron. 36:767–776.Peters F, Collette F, Degueldre C, Sterpenich V, Majerus S, Salmon E. 2009. The neural correlates of verbal

short-term memory in Alzheimer’s disease: an fMRI study. Brain. 132:1833–1846.Pihlajamäki M, DePeau K. 2008. Impaired medial temporal repetition suppression is related to failure of

parietal deactivation in Alzheimer disease. Am J Geriatr Psychiatry. 16:283–292.Pihlajamäki M, Sperling RA. 2009. Functional MRI assessment of task-induced deactivation of the default

mode network in Alzheimer’s disease and at-risk older individuals. Behav Neurol. 21:77–91.Raichle ME, Macleod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. 2001. A default mode of brain

function. Proc Natl Acad Sci U S A. 98:676–682.Rémy F, Mirrashed F, Campbell B, Richter W. 2005. Verbal episodic memory impairment in Alzheimer’s

disease: A combined structural and functional MRI study. Neuroimage. 25:253–266.Rombouts SARB, Goekoop R, Stam CJ, Barkhof F, Scheltens P. 2005. Delayed rather than decreased BOLD

response as a marker for early Alzheimer’s disease. Neuroimage. 26:1078–1085.Scahill RI, Schott JM, Stevens JM, Rossor MN, Fox NC. 2002. Mapping the evolution of regional atrophy in

Alzheimer’s disease: unbiased analysis of fluid-registered serial MRI. Proc Natl Acad Sci U S A. 99:4703–4707.

Seeley WW, Crawford RK, Zhou J, Miller BL, Greicius MD. 2009. Neurodegenerative diseases target large-scale human brain networks. Neuron. 62:42–52.

Shomstein S, Yantis S. 2006. Parietal cortex mediates voluntary control of spatial and nonspatial auditory attention. J Neurosci. 26:435–439.

Shulman GL, Fiez JA, Corbetta M, Buckner RL, Miezin FM, Raichle ME, Petersen SE. 1997. Common Blood Flow Changes across Visual Tasks: II. Decreases in Cerebral Cortex. J Cogn Neurosci. 9:648–663.

22

Sperling RA, Bates JF, Chua EF, Cocchiarella AJ, Rentz DM, Rosen BR, Schacter DL, Albert MS. 2003. fMRI studies of associative encoding in young and elderly controls and mild Alzheimer’s disease. J Neurol Neurosurg Psychiatry. 74:44–50.

Sperling RA, Dickerson BC, Pihlajamaki M, Vannini P, LaViolette PS, Vitolo O V, Hedden T, Becker JA, Rentz DM, Selkoe DJ, Johnson K a. 2010. Functional alterations in memory networks in early Alzheimer’s disease. Neuromolecular Med. 12:27–43.

Spreng RN, Grady CL. 2010. Patterns of brain activity supporting autobiographical memory, prospection, and theory of mind, and their relationship to the default mode network. J Cogn Neurosci. 22:1112–1123.

Strouse A, Hall JW, Burger MC. 1995. Central Auditory Processing in Alzheimer ’s Disease. Ear Hear. 230–238.

Thiyagesh SN, Farrow TFD, Parks RW, Accosta-Mesa H, Hunter MD, Young C, Wilkinson ID, Woodruff PWR. 2010. Treatment Effects of Therapeutic Cholinesterase Inhibitors on Visuospatial Processing in Alzheimer’s Disease: A Longitudinal Functional MRI Study. Dement Geriatr Cogn Disord. 29:176–188.

Vogt BA, Laureys S. 2005. Posterior cingulate, precuneal and retrosplenial cortices: Cytology and components of the neural network correlates of consciousness. Prog Brain Res. 150:205–217.

Warren JD, Fletcher PD, Golden HL. 2012. The paradox of syndromic diversity in Alzheimer disease. Nat Rev Neurol. 8:451–464.

Warren JD, Griffiths TD. 2003. Distinct mechanisms for processing spatial sequences and pitch sequences in the human auditory brain. J Neurosci. 23:5799–5804.

Warren JE, Wise RJS, Warren JD. 2005. Sounds do-able: auditory-motor transformations and the posterior temporal plane. Trends Neurosci. 28:636–643.

Warrington EK. 1984. Recogntion memory test. Windsor: NFER-Nelson.Warrington EK. 1996. The Camden Memory Test Battery. Hove: Psychology Press.Warrington EK, James M. 1991. The visual object and space perception battery. Bury St Edmunds: Thames

Valley Test Company.Wechsler D. 1987. Wechsler memory scale: Revised. San Antonio, TX: The Psychological Corporation.Wechsler D. 1997. Wechsler memory scale: WMS-III. San Antonio, TX: The Psychological Corporation,

Harcourt Brace.Wechsler D. 1999. Wechsler abbreviated scale of intelligence: WASI. San Antonio, TX: The Psychological

Corporation, Harcourt Brace.Weeks RA, Aziz-Sultan A, Bushara KO, Tian B, Wessinger CM, Dang N, Rauschecker JP, Hallett M. 1999. A

PET study of human auditory spatial processing. Neurosci Lett. 262:155–158.Weiskopf N, Hutton C, Josephs O, Deichmann R. 2006. Optimal EPI parameters for reduction of

susceptibility-induced BOLD sensitivity losses: a whole-brain analysis at 3 T and 1.5 T. Neuroimage. 33:493–504.

Wightman FL, Kistler DJ. 1989a. Headphone simulation of free-field listening. I: Stimulus synthesis. J Acoust Soc Am. 85:858–867.

Wightman FL, Kistler DJ. 1989b. Headphone simulation of free-field listening. II: Psychophysical validation. J Acoust Soc Am. 85:868–878.

Wong PCM, Jin JXJ, Gunasekera GMG, Abel R, Lee ER, Dhar S. 2009. Aging and cortical mechanisms of speech perception in noise. Neuropsychologia. 47:693–703.

Wong PCM, Uppunda AK, Parrish TB, Dhar S. 2008. Cortical mechanisms of speech perception in noise. J Speech Lang Hear Res. 51:1026–1041.

Xie C, Bai F, Yu H, Shi Y, Yuan Y, Chen G, Li W, Chen G, Zhang Z, Li SJ. 2012. Abnormal insula functional network is associated with episodic memory decline in amnestic mild cognitive impairment. Neuroimage. 63:320–327.

Zhou J, Seeley WW. 2014. Network dysfunction in Alzheimer’s disease and frontotemporal dementia: implications for psychiatry. Biol Psychiatry. 75:565–573.

Zimmer U, Lewald J, Erb M, Karnath H-O. 2006. Processing of auditory spatial cues in human cortex: an fMRI study. Neuropsychologia. 44:454–461.

23

Zündorf IC, Lewald J, Karnath H-O. 2013. Neural correlates of sound localization in complex acoustic environments. PLoS One. 8:e64259.

Zvyagintsev M, Clemens B, Chechko N, Mathiak KAK, Sack AT. 2013. Brain networks underlying mental imagery of auditory and visual information. Eur J Neurosci. 37:1421–1434.

Table 1. Demographic and neuropsychological characteristics of participant groups

Mean

(standard

deviation in

parentheses)

performance

scores are

shown

unless

otherwise

indicated.

Maximum

scores on

24

Characteristic Healthy controls ADGeneral demographic and clinicalNo. (m:f) 8:8 8:6Age (yrs) 70.1 (5.0) 69.8 (6.3)Handedness (R:L) 15:1 13:1Education (yrs) 16.0 (2.3) 13.3 (3.4)*MMSE (/30) 29 (1.1) 20 (5.1)Symptom duration (yrs) - 5.8 (2.0)General neuropsychological assessmentGeneral intellect: IQWASI verbal IQ 120 (8.9) 94 (17.2)*WASI performance IQ 121 (15.7) 93 (22.2)*NART estimated premorbid IQ 122 (5.5) 108 (15.6)*Episodic memoryRMT words (/50) 47 (2.2) 31 (7.4)*RMT faces (/50) 43 (4.2) 34 (6.9)*Camden PAL (/24) 21 (2.5) 3.4 (3.9)*Executive skillsWASI Block Design (/71) 43 (16.0) 19 (14.0)*WASI Matrices (/32) 28 (12.5) 13 (8.4)*WMS-R digit span forward (/12) 8.6 (1.9) 6.6 (1.7)*WMS-R digit span reverse (/12) 7.2(2.2) 4.7 (1.8)*WMS-III spatial span forward (/16) 6.8 (1.7) 5.1 (2.2)WMS-III spatial span reverse (/16) 6.9 (1.2) 3.4 (2.2)*D-KEFS Stroop colour (s)a 31 (7.3) 53 (21.0)*D-KEFS Stroop word (s)a 21 (4.2) 35 (18.1)*D-KEFS Stroop interference (s)a 65 (18.1) 103 (47.9)*Letter fluency (F: total) 17 (6.0) 9 (4.9)*Category fluency (animals: total) 21 (5.1) 11 (5.0)*Trails A (s)b 34 (10.7) 70 (50.3)*Trails B (s)c 78 (20.1) 196 (73.7)*WAIS-R Digit Symbol (total)d 52 (10.5) 26 (15.4)*Language skillsWASI Vocabulary (/80) 70 (4.6) 52 (13.7)*WASI Similarities (/48) 40 (6.9) 24 (12.4)*GNT (/30) 26 (2.0) 14 (7.8)*BPVS (/150) 147 (1.9) 135 (21.4)*NART (/50) 43 (4.5) 34 (10.7)*Posterior cortical skillsGDA (/24) 16 (4.2) 6 (6.2)*VOSP Object Decision (/20) 18 (2.2) 15 (3.7)*VOSP Dot Counting (/10) 9.9 (0.3) 8.6 (1.9)*Post-scan behavioural tasksAuditory spatial change detection (/20) 18.6(1.3) 14.9(3.6)*Pitch change detection (/20) 17.9(2.4) 15.6(3.6)*

neuropsychological tests are shown in parentheses. Results in bold indicate mean score < 5th

percentile for age norms (not available for BPVS and letter fluency); *significantly different from

control group (p < 0.05); a 13 patients completed this task; b 12 patients completed this subtest;

c 9 patients completed this subtest; d 11 patients completed this task. BPVS, British Picture

Vocabulary Scale (Dunn et al. 1982); D-KEFS, Delis Kaplan Executive System (Delis et al. 2001);

GDA, Graded Difficulty Arithmetic (Jackson and Warrington 1986); GNT, Graded Naming Test

(McKenna and Warrington 1983); NART, National Adult Reading Test (Nelson 1982); PAL, Paired

Associates Learning (Warrington 1996); RMT, Recognition Memory Test (Warrington 1984);

VOSP, Visual Object and Spatial Perception Battery (Warrington and James 1991); WASI,

Wechsler Abbreviated Scale of Intelligence (Wechsler 1999); WMS-R, Wechsler Memory Scale-

Revised (Wechsler 1987); WMS-III, Wechsler Memory Scale 3rd edition (Wechsler 1997).

25

Table 2. Summary of fMRI data for auditory contrasts of interest in participant groups

Group Contrast Region Side cluster(voxels)

Peak (mm) t-value p-valuex y z

HEALTHY CONTROLS

Sound > silencea HG/STG L 4236 -51 -15 1 21.51 <0.001HG/STG R 2704 58 -27 12 9.96 <0.001

Changing > fixed pitchb Anterior STG/STS R 477 59 2 -3 7.14 0.003Changing > fixed locationc PT/posterior STG L 933 -39 -37 15 8.69 0.001

PT/posterior STG R 584 66 -24 6 7.64 0.002Posterior cingulate cortex L 318 0 -48 34 6.29 0.016Posterior cingulate cortex R 109 2 -46 36 5.96 0.025

Changing pitch vs. changing locationd Anterior STG/STS L 53 -63 -12 4 6.34 0.008AD PATIENTS Sound > silencea HG/STG L 3301 -56 -10 -2 14.72 <0.001

HG/STG R 2007 48 -16 4 10.18 <0.001Changing pitch vs. changing locationd Posterior insula R 51 36 -16 7 7.52 0.005

CONTROLS >AD Changing > fixed locationc Posterior cingulate cortex L 95 0 -48 34 4.51 0.049Posterior cingulate cortex R 56 2 -48 34 4.51 0.049

AD > CONTROLS Changing pitch vs. changing locationd Posterior insula R 66 36 -16 9 4.77 0.016

Regional brain activations for contrasts between auditory conditions of interest within each participant group and between groups are shown; all

associations significant at peak-level threshold p < 0.05FWE corrected for multiple voxel-wise comparisons within pre-specified anatomical regions in

clusters >50 voxels in size are presented. Contrasts were composed as coded by superscripts: a, [(PfSf + PfSc + PcSf + PcSc) – silence]; b, [(PcSc + PcSf)

– (PfSc + PfSf)]; c, [(PcSc + PfSc) – (PcSf + PfSf)]; d, [(PcSc – PcSf) – (PfSc – PfSf)]. Conditions: PfSf = fixed pitch, fixed auditory spatial location; PcSf =

changing pitch, fixed spatial location; PfSc = fixed pitch, changing spatial location; PcSc = changing pitch, changing spatial location. AD, Alzheimer’s

disease; HG, Heschl’s gyrus; PT, planum temporale; STG/S, superior temporal gyrus/sulcus.

26